1,721,100 research outputs found
Valutazione dell’accuratezza statica dei convertitori analogico-digitale ad elevata risoluzione
No full textIl lavoro presenta un sistema realizzato su PC in grado di misurare e classificare gli errori di conversione in ADC commercialmente disponibili, permettendo di valutare il degrado delle prestazioni rispetto al comportamento ideale.
Vengono riportati infine i risultati ottenuti dall'analisi di due componenti commerciali a 14 e 16 bit
OpenMP and CUDA Simulations of Sella Zerbino Dam Break on Unstructured Grids
No full textThis paper presents two 2D dam break parallelized models based on shallow water equations (SWE) written in conservative form. The models were implemented exploiting multicore PC systems and graphics processor unit (GPU) architectures under the OpenMP and the NVIDIATM’s compute unified device architecture (CUDA) frameworks. The mathematical model is solved using a finite-volume technique on an unstructured grid, with Roe’s approximate Riemann solver, a first-order upwind scheme. The upwind treatment of the source terms is implemented. A technique to cope with a wetting-drying advance front is adopted, together with the inclusion of the influence of source terms in the stability constraint in order to prevent negative water depths at the dry fronts. The proposed model is first applied to a laboratory test and then to a real dam break that occurred in Italy in 1935. Results on different grid sizes are compared to show the computing efficiency between the original sequential model and the parallelized models
A parallel neural processor for real-time applications
No full textIn this paper we try to identify the most promising way to execute the training and the testing of a Multi Layer Perceptron neural network, by considering the use of a Matlab implementation on a PC and an embedded architecture based either on a general purpose multiprocessor or on a dedicated device. We compared the performance obtained from these three different approaches by first implementing a classification problem of recognising the region of the space to which randomly extracted points belong to and then facing a biomedical signal fitting problem. In both applications the superiority of the dedicated chip in terms of performance was evident. Moreover the dedicated chip uses a training technique, known as Reactive Tabu Search algorithm, which is often more efficient than the Back Propagation approach used in the other solutions
A Parallel Processor for Neural Networks
No full textWe present two different algorithms implemented through neural networks on a multiprocessor device. The parallel single-chip TI TMS32C80 Multimedia Video Processor (MVP). The goal of this experimentation is to test, on real problems, the performance of this powerful unit made up by one Master Risc Processor and by four Slave Digital Signal Processors (DSPs) and to evaluate its suitability to neural network applications. The first problem implemented is a typical classification algorithm in which the network recognises which points belong to different regions inside a 2D space. The second problem is more computationally heavy and consists of a network able to recognise `handwritten' digits. The parallel version of the first algorithm, was also tested on a commercially available supercomputer
FPGA based accelerators for computing intensive applications
No full textThis was a invited talk giving a brief overview about accelerator
Parallel Monte Carlo Simulations of Dipolar Systems: A New Approach to Accelerate the Potential Evaluation
No full textThe Monte Carlo-Metropolis (MM) and Molecular Dynamics (MD) techniques are among the most popular approaches to describe interacting particle systems [1]. While the former evaluates the potential arising from the particle interactions by identifying a number of statistically significative configurations, the latter requires the computation of the deterministic force field in the system. In both cases a quadratic relationship among the computing time and the number N of particles is established (O(N2)). The computational load is very high, since simulations consist of a high number of cycles to achieve a meaningful description of the physical system examined. In the physical systems we considered, periodic boundary conditions and minimum image conventions are applied, together with the Ewald method for summing the interactions between a particle and all the periodic images of the other. Even by choosing the parameters of the method in order to obtain a fixed accuracy and to minimize the computation time, however, it scales with O(N3/2), but remains still high [2][3]. Thus, due to the strong dependence of the computing time on the number of particles, only long simulations are able to describe complex systems.
A useful solution can be found in the use of look-up tables, where the potentials or forces are evaluated for particles located on fixed points of a 2D or 3D grid, at the beginning of the simulation. Afterwards, the potentials or force fields due to particles, not located on the grid, are more quickly obtained by interpolation of the originally tabulated values, without losing accuracy.
MM Simulations have been carried out, yielding encouraging results. Moreover, the speed-up achieved can be even greater if the algorithm is parallelised on a multiprocessor system. The overall speed-up achieved reaches a few thousands for a system with 1000 dipoles and it is well scalable with the dipole number. This makes feasible simulations which otherwise could take unacceptable execution time
A correlator for light-scattering experiments
No full textLight scattering is frequently used to examine the random motion of physical system particles. By evaluating the autocorrelation function of the acquired signal, structural informations of macromolecular solutions can be obtained.
We here present an instrument able to calculate the correlation function in real time. The instrument features high precision capabilities and a wide range of selectable delays. A Digital Signal Processor (DSP) is charged with the calculations, while a PC permits the user the management of the global instrument and the visualization and the storing of the evaluated correlation
Custom FPGA Processing for Real-Time Fetal ECG Extraction and Identification
Full text linkMonitoring the fetal cardiac activity during pregnancy is of crucial importance for evaluating fetus health. However, there is a lack of automatic and reliable methods for Fetal ECG (FECG) monitoring that can perform this elaboration in real-time. In this paper, we present a hardware architecture, implemented on the Altera Stratix V FPGA, capable of separating the FECG from the maternal ECG and to correctly identify it. We evaluated our system using both synthetic and real tracks acquired from patients beyond the 20th pregnancy week.
This work is part of a project aiming at developing a portable system for FECG continuous real-time monitoring. Its characteristics of reduced power consumption, real-time processing capability and reduced size make it suitable to be embedded in the overall system, that is the first proposed exploiting Blind Source Separation with this technology, to the best of our knowledge
A Hybrid CPU–GPU Real-Time Hyperspectral Unmixing Chain
Full text linkHyperspectral images are used in different applications in Earth and space science, and many of these applications exhibit real- or near real-time constraints. A problem when analyzing hyperspectral images is that their spatial resolution is generally not enough to separate different spectrally pure constituents (endmembers); as a result, several of them can be found in the same pixel. Spectral unmixing is an important technique for hyperspectral data exploitation, aimed at finding the spectral signatures of the endmembers and their associated abundance fractions. The development of techniques able to provide unmixing results in real-time is a long desired goal in the hyperspectral imaging community. In this paper, we describe a real-time hyperspectral unmixing chain based on three main steps: 1) estimation of the number of endmembers using the hyperspectral subspace identification with minimum error (HySime); 2) estimation of the spectral signatures of the endmembers using the vertex component analysis (VCA); and 3) unconstrained abundance estimation. We have developed new parallel implementations of the aforementioned algorithms and assembled them in a fully operative real-time unmixing chain using graphics processing units (GPUs), exploiting NVIDIA's compute unified device architecture (CUDA) and its basic linear algebra subroutines (CuBLAS) library, as well as OpenMP and BLAS for multicore parallelization. As a result, our real-time chain exploits both CPU (multicore) and GPU paradigms in the optimization. Our experiments reveal that this hybrid GPU-CPU parallel implementation fully meets real-time constraints in hyperspectral imaging applications
- …
